Hand recognition is a recent active area of research in the computer vision for the purpose of Human -Computer Interaction. This paper mainly concentrates on Tamil sign alphabets (TSL) into speech which could be helpful for deaf-dumb people. In this paper, a set of 32 (2 5 ) combinations of binary number sign images are introduced to propose a system to recognize Tamil sign alphabets. These Tamil alphabets have 12 vowels, 18 consonants, and one Aayutha Ezhuthu. The proposed system is based on four main stages: Pre-processing method, Training phase, Sign detection, and Conversion of Binary to voice. The binary sign images are loaded at a run time or static as 310 images which are taken ten times in different distances at the same position. The five fingertip positions represent ('1' or '0') and are identified by using image processing techniques with proposed right hand palm angular-based analysis. Then, the binary values are assigned to the corresponding Tamil letters and voice. The experiments were performed with ten different signer palms and the results demonstrated that the system could successfully recognize Tamil sign alphabets with better accuracy with 99.35% of static images and 98.36% of dynamic images (runtime).